Comparative Solution Project

I noticed a large value for dispersivity (10 m) which actually lead to a noticeable impact on the solubility of CO2 at reservoir scale using OPM. I wonder where that value come from. Is it typical for some of the Norwegian aquifers?

Hi Alberto, 

Thanks for your question.

The dispersivity was chosen based on a couple of different considerations:

1) We wanted to be more-or-less consistent with the standard literature understanding of macro-dispersivity in porous media, which postulates that the dispersion is length-scale dependent. In my understanding of this topic, the most natural length scale for the CSP, is the scale below which the material is described as homogeneous, which in this case is essentially the characteristic facies thickness (100-500 meters).  This suggests a dispersivity on the order of decades of meters, according to e.g. the summary of relevant data provided in Larry Lake's book (and also classical groundwater books). 

2) Since several simulators do not include (physical) dispersion in their multi-phase flow modules, we also wanted to choose a dispersion coefficient which was small enough that it would be more-or-less dominated by the numerical dispersion of the standard low-order discretizations when employed on the reporting grid. 

3) Finally, since dispersion directly impacts the characteristic width/cross-section of density-driven fingers, it is one of the key factors that determines the scale which must be resolved to obtain a grid-converged solution for the problem. As such, we also did not want to set the dispersion too low, since this would make the task of obtaining high-resolution simulations of the CSP even harder. 

In view of the above, 10m seemed like a reasonable compromise value, both in terms of the physics and the numerics.

All the best, 

Jan